Overview

This report provides exploratory and preliminary correlation analyses between sex hormones and T2 unretouched, raw hippocampal subfield data from the Healthy Aging Study sample (subjects 333 - 437).

Cleaning data steps included:

  1. Filter participants that have T2 subfield data from scan
  2. Select sex hormone measures from when scan was conducted
  3. Correct left and right volumes with TIV from VBM
  4. Average left and right volumes to get an average measure for each subfield region

Sample size:

  • Men = 24

  • Women = 63 (24 pre, 20 peri, 19 post)

Structural Distributions

Shapiro-Wilk’s test reveal that CA23 for men, and SUB overall and SUB for women were not normally distributed. Hormone and subfield associations for these subfields will be using spearman rank correlation

Hormone Distributions

Shapiro-Wilk test reveals that estradiol, FSH, and testosterone for all and for women subgroup distributions were still significantly different from normal after log transformation, so Spearman rank correlations will be applied to these hormone and subfield associations.

Reproductive Status

For women, hormones will be log transformed to allow for Pearson correlations to run in the subsequent analyses (by reproductive stage). Estradiol, FSH, and testosterone correlations will be Spearman rank correlations for women.

Men and Women

For any correlation analyses between women and men, hormones will also be log transformed for Pearson correlations. Testosterone associations will be using Spearman rank.

Men Only

For just men, log transformation will also be applied in the following Pearson correlations. Testosterone correlations among men will be using Spearman rank.

T2 Subfield Volume

CA1

Structural Comparison

repo_status mean_ca1 std_ca1
Pre 0.594 0.072
Peri 0.601 0.072
Post 0.564 0.067
sex mean_ca1 std_ca1
Female 0.587 0.071
Male 0.540 0.056

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.0151012 0.0075506 1.520983 0.2268011
Residuals 60 0.2978570 0.0049643 NA NA

ANOVA results comparing CA1 subfield by reproductive status was not significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_ca1 by sex
## t = 3.216, df = 52.242, p-value = 0.002232
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  0.01761859 0.07607124
## sample estimates:
## mean in group Female   mean in group Male 
##            0.5872078            0.5403629

T-test results comparing CA1 subfield by sex was significant.

Hormone Correlation by Reproductive Status

Correlations for estradiol, FSH, and testosterone were Spearman rank. All others were Pearson.

Hormone Correlation by Men

Correlations are all Pearson

CA2/3

Structural Comparison

repo_status mean_ca23 std_ca23
Pre 0.252 0.048
Peri 0.247 0.040
Post 0.239 0.042
sex mean_ca23 std_ca23
Female 0.247 0.044
Male 0.223 0.037

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.0019035 0.0009517 0.493987 0.6126484
Residuals 60 0.1155978 0.0019266 NA NA

ANOVA results comparing CA23 subfield by reproductive status was not significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_ca23 by sex
## t = 2.5392, df = 48.975, p-value = 0.01434
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  0.004920838 0.042268912
## sample estimates:
## mean in group Female   mean in group Male 
##            0.2467289            0.2231340

T-test results comparing CA23 subfield by sex was significant.

Hormone Correlation by Reproductive Status

Correlations for estradiol, FSH, and testosterone were Spearman rank. All others were Pearson.

Hormone Correlation by Men

Correlations are all Spearman.

DG

Structural Comparison

repo_status mean_dg std_dg
Pre 0.632 0.080
Peri 0.619 0.072
Post 0.604 0.070
sex mean_dg std_dg
Female 0.619 0.074
Male 0.583 0.072

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.0083941 0.0041971 0.750208 0.4766459
Residuals 60 0.3356715 0.0055945 NA NA

ANOVA results comparing DG subfield by reproductive status was not significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_dg by sex
## t = 2.0854, df = 43.206, p-value = 0.04298
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  0.00119681 0.07118450
## sample estimates:
## mean in group Female   mean in group Male 
##            0.6192800            0.5830894

T-test results comparing DG subfield by sex was significant.

Hormone Correlation by Reproductive Status

Correlations for estradiol, FSH, and testosterone were Spearman rank. All others were Pearson.

Hormone Correlation by Men

All correlations are Pearson.

ERC

Structural Comparison

repo_status mean_erc std_erc
Pre 0.480 0.080
Peri 0.485 0.070
Post 0.427 0.056
sex mean_erc std_erc
Female 0.466 0.074
Male 0.414 0.055

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.0407582 0.0203791 4.123448 0.0209918
Residuals 60 0.2965350 0.0049423 NA NA

ANOVA results comparing ERC subfield by reproductive status was significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_erc by sex
## t = 3.5175, df = 55.723, p-value = 0.0008746
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  0.02204370 0.08038099
## sample estimates:
## mean in group Female   mean in group Male 
##            0.4655728            0.4143605

T-test results comparing ERC subfield by sex was significant.

Hormone Correlation by Reproductive Status

Correlations for estradiol, FSH, and testosterone were Spearman rank. All others were Pearson.

Hormone Correlation by Men

All correlations are Pearson.

PHC

Structural Comparison

repo_status mean_phc std_phc
Pre 1.707 0.183
Peri 1.638 0.202
Post 1.673 0.225
sex mean_phc std_phc
Female 1.675 0.201
Male 1.557 0.219

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.0516381 0.0258191 0.6318458 0.5351094
Residuals 60 2.4517754 0.0408629 NA NA

ANOVA results comparing PHC subfield by reproductive status was not significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_phc by sex
## t = 2.2998, df = 38.699, p-value = 0.02694
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  0.01419269 0.22180438
## sample estimates:
## mean in group Female   mean in group Male 
##             1.674716             1.556717

T-test results comparing PHC subfield by sex was significant.

Hormone Correlation by Reproductive Status

Correlations for estradiol, FSH, and testosterone were Spearman rank. All others were Pearson.

Hormone Correlation by Men

All correlations are Pearson.

PRC

Structural Comparison

repo_status mean_prc std_prc
Pre 1.627 0.375
Peri 1.602 0.355
Post 1.740 0.373
sex mean_prc std_prc
Female 1.653 0.367
Male 1.607 0.384

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.211709 0.1058545 0.7813027 0.4624106
Residuals 60 8.129078 0.1354846 NA NA

ANOVA results comparing PRC subfield by reproductive status was not significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_prc by sex
## t = 0.50652, df = 39.953, p-value = 0.6153
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  -0.1378815  0.2301021
## sample estimates:
## mean in group Female   mean in group Male 
##             1.652974             1.606863

T-test results comparing PRC subfield by sex was not significant.

Hormone Correlation by Reproductive Status

Correlations for estradiol, FSH, and testosterone were Spearman rank. All others were Pearson.

Hormone Correlation by Men

All correlations are Pearson.

SUB

Structural Comparison

repo_status mean_sub std_sub
Pre 0.381 0.052
Peri 0.379 0.037
Post 0.358 0.049
sex mean_sub std_sub
Female 0.373 0.047
Male 0.346 0.036

Means and standard deviations of the subfield are displayed by reproductive status and by sex

Df Sum Sq Mean Sq F value Pr(>F)
repo_status 2 0.0060737 0.0030368 1.392181 0.2564443
Residuals 60 0.1308814 0.0021814 NA NA

ANOVA results comparing SUB subfield by reproductive status was not significant.

## 
##  Welch Two Sample t-test
## 
## data:  avg_sub by sex
## t = 2.8303, df = 53.988, p-value = 0.006517
## alternative hypothesis: true difference in means between group Female and group Male is not equal to 0
## 95 percent confidence interval:
##  0.007799812 0.045688732
## sample estimates:
## mean in group Female   mean in group Male 
##            0.3732318            0.3464875

T-test results comparing SUB subfield by sex was significant.

Hormone Correlation by Reproductive Status

All correlations are Spearman.

Hormone Correlation by Men

All correlations are Pearson

Post-Menopausal Women: E, P, FSH

Correlations with E, FSH, and testosterone are Spearman rank. Correlations with SUB are all Spearman rank. All others are Pearson correlations.

Summary

  • All hormones were log transformed first to try to normalize distributions. Next, all correlations with estradiol, FSH, and testosterone for women and by reproductive stage were Spearman rank correlations. Correlations for SUB for women were also all Spearman rank. Correlations for men between hormones and CA23 were also all Spearman rank. All other correlations are Pearson correlations.

  • Subregion volumes for women and men were generally normally distributed, except for CA23 in men and SUB in women. Any associations/correlations ran with these groups for these subfields were Spearman rank correlation. There appears to be two women whose SUB volumes appear to be outliers. Hormones were generally right skewed, so log transformations were applied with some still not being corrected fully - so Spearman correlations for certain associations were used.

  • No significant trends seen in correlations for DHEAS.

  • No significant trends seen in correlations for DG.

  • No significant trends seen in correlations for men.

  • ANOVA results comparing subfields by reproductive stage were not significant for almost all subfields. ANOVA results for ERC was significant by reproductive stage (p < 0.05).

  • T-test results comparing subfields by sex were all significant (at least p < 0.05) except for PRC.

  • For estradiol, it was significantly negatively correlated with CA1 for peris (p = 0.036, Spearman) and with SUB for pres (p = 0.04, Spearman), and marginally negatively correlated with CA23 for pres (p = 0.054, Spearman) and peris (p = 0.068, Spearman).

  • For progesterone, it was significantly positively correlated with CA1 for posts (p = 0.033, Pearson) and with ERC for posts (p = 0.015, Pearson).

  • For FSH, it was significantly negatively correlated with ERC for posts (p < 0.01, Spearman) and with SUB for posts (p = 0.021, Spearman). PHC volume and FSH were also significantly negatively correlated among women as a group (p = 0.041, Spearman). FSH is marginally negatively correlated with CA1 for posts (p = 0.086, Spearman).

  • For testosterone, it was marginally negatively correlated with ERC for posts (p = 0.083, Spearman) and marginally positively correlated with PHC for pres (p = 0.085, Spearman).

  • For SHBG, it was significantly negatively correlated with ERC for posts (p < 0.01, Pearson). It was also negatively correlated with PRC among women as a group (p = 0.028, Pearson) and marginally negatively correlated with PRC for pres (p = 0.057). SHBG was also marginally negatively correlated with SUB for posts (p = 0.085, Spearman).